Changeset 802
- Timestamp:
- 10/10/08 14:49:46 (17 months ago)
- Location:
- docs/Lowthane
- Files:
-
- 5 modified
-
design.tex (modified) (8 diffs)
-
evaluation.tex (modified) (1 diff)
-
lowthane.tex (modified) (2 diffs)
-
related.tex (modified) (2 diffs)
-
tossim.tex (modified) (1 diff)
Legend:
- Unmodified
- Added
- Removed
-
docs/Lowthane/design.tex
r799 r802 150 150 \begin{figure} [h] 151 151 \centering 152 \includegraphics[width=0. 8\columnwidth]{figs/topology_collection.png}152 \includegraphics[width=0.5\columnwidth]{figs/topology_collection.png} 153 153 \caption{An example collected topology when only one link is reported by each 154 154 node. We report four links per node.}\label{fig:topology} … … 262 262 \begin{figure} [h] 263 263 \centering 264 \includegraphics[width=0. 8\columnwidth]{figs/routing.png}264 \includegraphics[width=0.5\columnwidth]{figs/routing.png} 265 265 \caption{The effect of the route install primitive.}\label{fig:routing} 266 266 \end{figure} … … 322 322 %% cause thrashing, which leads to degraded performance. We explore this in 323 323 %% detail under various conditions and workloads in section~\ref{sec:eval}. 324 \subsection{Forwarding Discipline} 324 \section{Our Hybrid Approach} 325 325 326 {\bf Decision:} The default route as used as a backup in the case where installed state 326 327 is invalid. … … 427 428 \subsection{Shortcomings of Centralized Protocols} 428 429 429 Introduction of a centralized element of routing raises a family of concerns. We briefly examine the most common ones raised in the literature and anecdotally in the context of our setting and approach.\\ 430 \\ 430 Introduction of a centralized element of routing raises a family of concerns. We briefly examine the most common ones raised in the literature and anecdotally in the context of our setting and approach. 431 431 \noindent 432 432 {\bf Latency of sending packet through controller: }The latency incurred by … … 437 437 real-time scenarios in which latency is critical; in such cases an on-demand 438 438 solution like {\lowthane} may not be appropriate without quality-of-service 439 adaptations. \\440 \\ 439 adaptations. 440 441 441 \noindent 442 442 {\bf Controller becomes a single point of failure: }The failure of the … … 448 448 in the network and the number of controllers. We discuss this further in 449 449 section~\ref{sec:extensions}, but for simplicity focus on single-controller 450 networks until then. \\451 \\ 450 networks until then. 451 452 452 \noindent 453 453 {\bf A persistent back channel to the controller is needed: }The directed acyclic … … 459 459 to dynamic topological conditions, and also maintains multiple options for 460 460 additional reliability. Our results in section~\ref{sec:eval} demonstrate the 461 reliability of this back channel. \\462 \\ 461 reliability of this back channel. 462 463 463 \noindent 464 464 {\bf Difficult to maintain consistent global view of topology: }This … … 473 473 initial evidence that it is possible to react sufficiently quickly to network 474 474 changes. 475 \\ 476 \\ 475 477 476 478 477 \subsection{ROLL Requirements} -
docs/Lowthane/evaluation.tex
r796 r802 31 31 \begin{figure} [h] 32 32 \centering 33 \includegraphics[width=0. 8\columnwidth]{figs/Soda-4_2008-09-24.png}33 \includegraphics[width=0.5\columnwidth]{figs/Soda-4_2008-09-24.png} 34 34 \caption{79-node Smote Testbed at UC Berkeley}\label{fig:smote} 35 35 \end{figure} -
docs/Lowthane/lowthane.tex
r796 r802 39 39 high control overhead, or poor response to loss. We propose a new design for 40 40 routing in this class of networks which takes advantage of typical device 41 heterogeneity and traffic patterns in this environ ement. We present a hybrid41 heterogeneity and traffic patterns in this environment. We present a hybrid 42 42 between a centralized and distributed routing algorithm, which uses a 43 43 distributed algorithm to provide reliable routes to a small set of 44 44 controllers. Given that the predominant traffic patterns in these networks 45 45 are collection and dissemination, these routes optimize for the common case. 46 The focus of this work is to evalu te the potential for further optimization of46 The focus of this work is to evaluate the potential for further optimization of 47 47 point-to-point traffic within the network through the use of a centralized 48 48 solution. In our design, controllers maintain a view of the network topology as … … 55 55 performance, but to present preliminary results showing that it performs well 56 56 in a range of expected situations, thus demonstrating the design's viability. 57 Finally, we evalu te the state and traffic complexity of our design to57 Finally, we evaluate the state and traffic complexity of our design to 58 58 demonstrate that it meets requirements for the space. 59 59 -
docs/Lowthane/related.tex
r797 r802 37 37 single destination routing, and also affords no flexibility for dynamic 38 38 topology conditions (unless links completely break). TSMP, which is 39 incorporated in the WirelessHART standard is a well-established industry39 incorporated in the WirelessHART standard, is a well-established industry 40 40 approach that includes slotted time \& channel scheduling using the global 41 41 information, but little technical data is available for it. … … 65 65 66 66 Finally Hui's work explores extending the 67 internet architecture to L2Ns, and develops the algorithms necessary for67 Internet architecture to L2Ns, and develops the algorithms necessary for 68 68 default route formation in \cite{hui}. This work has been invaluable in 69 69 helping us to frame our architecture. -
docs/Lowthane/tossim.tex
r801 r802 37 37 \begin{figure}[h] 38 38 \centering 39 \includegraphics[width=. 8\columnwidth]{figs/collection.pdf}39 \includegraphics[width=.5\columnwidth]{figs/collection.pdf} 40 40 \caption{Our default route selection protocol compared with CTP in simulation 41 41 with a poor noise model, and to results on our testbed.}
